Forward collision avoidance systems have shown to be a particularly effective crash-avoidance technology. Multivehicle tracking capabilities play an important role in the real-world performance and effectiveness of such systems. In order to effectively and accurately track vehicles in a moving platform and in complicated road environments, the authors proposed a multi-vehicle tracking algorithm based on an improved particle filter. First, the authors used a vehicle disappearance detection and handling mechanism based on the normalised area of the minimum circumscribed rectangle of particle distributions. This mechanism is used to verify whether a new target is a vehicle and can also handle the vehicle exit during the tracking phase. Next, an improved particle filter-based framework, which includes a new process dynamical distribution, allowed for multi-vehicle tracking capabilities was used for vehicle tracking. Finally, an effective occlusion detection and handling mechanism was used to address the significant occlusion between vehicles. The combination of these added improvements in the algorithm results in the enhancement of the vehicle tracking rate in a variety of challenging conditions. Experimental tests carried out from different datasets show excellent performance in multi-vehicle tracking, in terms of accuracy in complex traffic situations and under different lighting conditions.
The current vehicle tracking algorithms cannot meet the requirements of high robustness in engineering application. A co-training algorithm based on on-line boosting for vehicle tracking is proposed. In this algorithm, first the vehicle region of interest is detected by vehicle-shadow feature and vehicle horizontal edge feature. Then the vehicle region of interest is verified by off-line classifiers which are learned from Haar feature and Adaboost algorithm. Finally, a co-training algorithm based on on-line boosting is used for further vehicle tracking, then the tracking window was reshaped according to the shadow of target vehicle. Experiments show that the proposed algorithm has high robustness and flexibility with good application prospects.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.